Session level network usage patterns of mobile handsets

Abstract

In this paper we utilize mobile handset-based measurements to examine smartphone generated mobile network usage. The focus is on user behavior analysis and network session is defined as the measure of network usage. In particular, we utilize the richness of handset-based data to understand the offline/background actions leading to the network usage. We make a distinction between human versus machine-generated network sessions, and examine the sessions in different semantic places of users, as well as, on different battery levels of the users' devices. Machine-generated sessions show differing properties compared to human-generated sessions, and also other contextual factors affect the properties of the network sessions. The results could be applied to more generalizable network node measurements to gain a more large-scale view on mobile network usage.